R Notebook

# code to load packages
library(tidyverse)
library(tidymodels)
library(knitr)
library(dplyr)
library(ggplot2)
library(patchwork)
library(corrplot)

#baseR

auto <- read_csv("data/autodatawg.csv")

New names:
• `notes` -> `notes...23`
• `notes` -> `notes...61`
• `` -> `...99`
Rows: 583 Columns: 123
── Column specification ────────────────────────────────────────────────────────
Delimiter: ","
chr (109): Collector, Higher taxon, Taxon, aagroup, miss mC, miss mT, miss  ...
dbl   (2): radius L, Ulna L
lgl  (12): m2-5 av, hpp2-5 av, hmp2-5 av, hdp2-5 av, notes...61, fpp2-5 av, ...

ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

#Set hms as numeric 
auto$hm1 <- as.numeric(as.character(auto$hm1))

Warning: NAs introduced by coercion

auto$hm2 <- as.numeric(as.character(auto$hm2))

Warning: NAs introduced by coercion

auto$hm3 <- as.numeric(as.character(auto$hm3))

Warning: NAs introduced by coercion

auto$hm4 <- as.numeric(as.character(auto$hm4))

Warning: NAs introduced by coercion

auto$hm5 <- as.numeric(as.character(auto$hm5))

Warning: NAs introduced by coercion

#Set hpps as numeric 
auto$hpp1 <- as.numeric(as.character(auto$hpp1))

Warning: NAs introduced by coercion

auto$hpp2 <- as.numeric(as.character(auto$hpp2))

Warning: NAs introduced by coercion

auto$hpp3 <- as.numeric(as.character(auto$hpp3))

Warning: NAs introduced by coercion

auto$hpp4 <- as.numeric(as.character(auto$hpp4))

Warning: NAs introduced by coercion

auto$hpp5 <- as.numeric(as.character(auto$hpp5))

Warning: NAs introduced by coercion

#remove NAs
autoclean <- auto %>%
  filter(complete.cases(hm1, hm2, hm3, hm4, hm5, hpp1, hpp2, hpp3, hpp4, hpp5))

#avg hand metatarsals for each higher taxa

hm_avg <- autoclean |>
  group_by(aagroup) |>
  summarize(average_hm1 = mean(hm1, na.rm = TRUE),
            average_hm2 = mean(hm2, na.rm = TRUE),
            average_hm3 = mean(hm3, na.rm = TRUE),
            average_hm4 = mean(hm4, na.rm = TRUE),
            average_hm5 = mean(hm5, na.rm = TRUE)
            )
print(hm_avg)

#avg proximal phalanges for each higher taxa

hpp_avg <- autoclean |>
  group_by(aagroup) |>
  summarize(average_hpp1 = mean(hpp1, na.rm = TRUE),
            average_hpp2 = mean(hpp2, na.rm = TRUE),
            average_hpp3 = mean(hpp3, na.rm = TRUE),
            average_hpp4 = mean(hpp4, na.rm = TRUE),
            average_hpp5 = mean(hpp5, na.rm = TRUE)
            )
print(hpp_avg)

#get total digit lengths 

autoclean$total_digit1 <- autoclean$hm1 + autoclean$hpp1
autoclean$total_digit2 <- autoclean$hm2 + autoclean$hpp2
autoclean$total_digit3 <- autoclean$hm3 + autoclean$hpp3
autoclean$total_digit4 <- autoclean$hm4 + autoclean$hpp4
autoclean$total_digit5 <- autoclean$hm5 + autoclean$hpp5

head(autoclean)

NA

#AVG DIGIT LENGTHS

avg_digit_lengths <- hm_avg |>
  full_join(hpp_avg, by = "aagroup") |>

  mutate(
      avg_digit1 = hm_avg$average_hm1 + hpp_avg$average_hpp1,
    avg_digit2 = hm_avg$average_hm2 + hpp_avg$average_hpp2,
    avg_digit3 = hm_avg$average_hm3 + hpp_avg$average_hpp3,
    avg_digit4 = hm_avg$average_hm4 + hpp_avg$average_hpp4,
    avg_digit5 = hm_avg$average_hm5 + hpp_avg$average_hpp5
    ) |>
    select(aagroup, avg_digit1, avg_digit2, avg_digit3, avg_digit4, avg_digit5)

avg_digit_lengths

#EXPLORING COVARIATION
# BIPLOTS FOR ALL DIGITS IN DATASET 

#to calc rsquared 
r_squared <- function(x, y) {
  model <- lm(y ~ x)
  summary(model)$r.squared
}

pairs <- list(
  c("total_digit1", "total_digit2"),
  c("total_digit1", "total_digit3"),
  c("total_digit1", "total_digit4"),
  c("total_digit1", "total_digit5"),
  c("total_digit2", "total_digit3"),
  c("total_digit2", "total_digit4"),
  c("total_digit2", "total_digit5"),
  c("total_digit3", "total_digit4"),
  c("total_digit3", "total_digit5"),
  c("total_digit4", "total_digit5")
)

for (pair in pairs) {
  x <- autoclean[[pair[1]]]
  y <- autoclean[[pair[2]]]
  r_squared <- calculate_r_squared(x, y)
  p <- ggplot(autoclean, aes_string(x = pair[1], y = pair[2])) +
    geom_point() +
    labs(x = pair[1], y = pair[2], title = paste("Biplot of", pair[1], "and", pair[2]), subtitle = paste("R^2 =", r_squared)) +
    theme_minimal()
  print(p)
}

Warning: `aes_string()` was deprecated in ggplot2 3.0.0.
ℹ Please use tidy evaluation idioms with `aes()`.
ℹ See also `vignette("ggplot2-in-packages")` for more information.
This warning is displayed once every 8 hours.
Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
generated.

#CORRELATION MATRIX 
correlation_matrix <- cor(autoclean[, c("total_digit1", "total_digit2", "total_digit3", "total_digit4", "total_digit5")])
print(correlation_matrix)

             total_digit1 total_digit2 total_digit3 total_digit4 total_digit5
total_digit1    1.0000000    0.9566183    0.9565192    0.9634344    0.9664044
total_digit2    0.9566183    1.0000000    0.9791062    0.9725032    0.9686648
total_digit3    0.9565192    0.9791062    1.0000000    0.9941848    0.9786560
total_digit4    0.9634344    0.9725032    0.9941848    1.0000000    0.9928006
total_digit5    0.9664044    0.9686648    0.9786560    0.9928006    1.0000000

corrplot(correlation_matrix, method = "color", addCoef.col = "black", tl.col = "black", tl.srt = 45)

# Function to calculate R^2
calc_r_squared <- function(x, y) {
  model <- lm(y ~ x)
  summary(model)$r.squared
}

# Create pairs of digit lengths
pairs <- list(
  c("total_digit1", "total_digit2"),
  c("total_digit1", "total_digit3"),
  c("total_digit1", "total_digit4"),
  c("total_digit1", "total_digit5"),
  c("total_digit2", "total_digit3"),
  c("total_digit2", "total_digit4"),
  c("total_digit2", "total_digit5"),
  c("total_digit3", "total_digit4"),
  c("total_digit3", "total_digit5"),
  c("total_digit4", "total_digit5")
)

# Get unique taxa
unique_taxa <- unique(autoclean$aagroup)

# for each higher taxa 
for (taxon in unique_taxa) {
  # filter for the current higher taxon
  taxa_data <- autoclean[autoclean$aagroup == taxon, ]
  
  # going thru each digit pair 
  for (pair in pairs) {
    x <- taxa_data[[pair[1]]]
    y <- taxa_data[[pair[2]]]
    
    # get r^2
    r_squared <- calc_r_squared(x, y)
    
    # actual plot
    p <- ggplot(taxa_data, aes_string(x = pair[1], y = pair[2])) +
      geom_point(color = "#0096FF", alpha = 0.6, size = 3) +
      labs(
        x = pair[1],
        y = pair[2],
        title = paste("Biplot of", pair[1], "and", pair[2], "for", taxon),
        subtitle = paste("R^2 =", round(r_squared, 3))
      ) +
      theme_minimal() +
      facet_wrap(~ `Sex/age`) +
      theme(
        plot.title = element_text(size = 12, face = "bold", color = "#0096FF"),
        plot.subtitle = element_text(size = 10, color = "black"),
        axis.title.x = element_text(size = 10, face = "bold"),
        axis.title.y = element_text(size = 10, face = "bold"),
        panel.background = element_rect(fill = "#FFF1F3"),  
        panel.grid.major = element_line(color = "white"), 
        panel.grid.minor = element_line(color = "white")
      )
    print(p)
  }
}

# Function to calculate R^2
calc_r_squared <- function(x, y) {
  model <- lm(y ~ x)
  summary(model)$r.squared
}

# Create pairs of digit lengths
pairs <- list(
  c("total_digit1", "total_digit2"),
  c("total_digit1", "total_digit3"),
  c("total_digit1", "total_digit4"),
  c("total_digit1", "total_digit5"),
  c("total_digit2", "total_digit3"),
  c("total_digit2", "total_digit4"),
  c("total_digit2", "total_digit5"),
  c("total_digit3", "total_digit4"),
  c("total_digit3", "total_digit5"),
  c("total_digit4", "total_digit5")
)

# Get unique taxa
unique_taxa <- unique(autoclean$aagroup)

# for each higher taxa 
for (taxon in unique_taxa) {
  # filter for the current higher taxon
  taxa_data <- autoclean[autoclean$aagroup == taxon, ]
 
  # going thru each digit pair 
  for (pair in pairs) {
    x <- taxa_data[[pair[1]]]
    y <- taxa_data[[pair[2]]]
    
    # get r^2
    r_squared <- calc_r_squared(x, y)
    
    # actual plot
    p <- ggplot(taxa_data, aes_string(x = pair[1], y = pair[2])) +
      geom_point(color = "#0096FF", alpha = 0.6, size = 3) +
      labs(
        x = pair[1],
        y = pair[2],
        title = paste("Biplot of", pair[1], "and", pair[2], "for", taxon),
        subtitle = paste("R^2 =", round(r_squared, 3))
      ) +
      theme_minimal() +
      theme(
        plot.title = element_text(size = 12, face = "bold", color = "#0096FF"),
        plot.subtitle = element_text(size = 10, color = "black"),
        axis.title.x = element_text(size = 10, face = "bold"),
        axis.title.y = element_text(size = 10, face = "bold"),
        panel.background = element_rect(fill = "#FFF1F3"),  
        panel.grid.major = element_line(color = "white"), 
        panel.grid.minor = element_line(color = "white")
      )
    print(p)
  }

}

NA

# Get unique taxa
unique_taxa <- unique(autoclean$aagroup)

# for each higher taxa 
for (taxon in unique_taxa) {
  # filter for the current higher taxon
  taxa_data <- autoclean[autoclean$aagroup == taxon, ]
  
  #empty matrix 
   r_squared_matrix <- matrix(NA, nrow = 5, ncol = 5)
    rownames(r_squared_matrix) <- colnames(r_squared_matrix) <- c("total_digit1", "total_digit2", "total_digit3", "total_digit4", "total_digit5")
    
 for (pair in pairs) {
    x <- taxa_data[[pair[1]]]
    y <- taxa_data[[pair[2]]]
    
    # get r^2
    r_squared <- calc_r_squared(x, y)
    
    # Store R^2 in the matrix
    r_squared_matrix[pair[1], pair[2]] <- r_squared
    r_squared_matrix[pair[2], pair[1]] <- r_squared
 }
      print(paste("R^2 matrix for taxon:", taxon))
    print(r_squared_matrix)
    
    # Plot R^2 matrix as a heatmap
    corrplot(r_squared_matrix, method = "color", addCoef.col = "black", tl.col = 
               "black", tl.srt = 45, title = paste("R^2 Matrix for", taxon), mar=c(0,0,1,0))
  
}

[1] "R^2 matrix for taxon: PAN"
             total_digit1 total_digit2 total_digit3 total_digit4 total_digit5
total_digit1           NA            0            0            0            0
total_digit2            0           NA            0            0            0
total_digit3            0            0           NA            0            0
total_digit4            0            0            0           NA            0
total_digit5            0            0            0            0           NA
[1] "R^2 matrix for taxon: OWM"
             total_digit1 total_digit2 total_digit3 total_digit4 total_digit5
total_digit1           NA    0.5894414    0.5515566    0.5649344    0.5957257
total_digit2    0.5894414           NA    0.9612788    0.9602187    0.9448672
total_digit3    0.5515566    0.9612788           NA    0.9960620    0.9811747
total_digit4    0.5649344    0.9602187    0.9960620           NA    0.9880566
total_digit5    0.5957257    0.9448672    0.9811747    0.9880566           NA

[1] "R^2 matrix for taxon: COLUGO"
             total_digit1 total_digit2 total_digit3 total_digit4 total_digit5
total_digit1           NA            0            0            0            0
total_digit2            0           NA            0            0            0
total_digit3            0            0           NA            0            0
total_digit4            0            0            0           NA            0
total_digit5            0            0            0            0           NA

[1] "R^2 matrix for taxon: NWM"
             total_digit1 total_digit2 total_digit3 total_digit4 total_digit5
total_digit1           NA    0.9646182    0.9528415    0.9295581    0.9203128
total_digit2    0.9646182           NA    0.9846387    0.9588601    0.9485791
total_digit3    0.9528415    0.9846387           NA    0.9862927    0.9679882
total_digit4    0.9295581    0.9588601    0.9862927           NA    0.9884083
total_digit5    0.9203128    0.9485791    0.9679882    0.9884083           NA

[1] "R^2 matrix for taxon: SCANDENTIA"
             total_digit1 total_digit2 total_digit3 total_digit4 total_digit5
total_digit1           NA    0.9243360    0.8907543    0.9102392    0.9646054
total_digit2    0.9243360           NA    0.9817450    0.9766877    0.9619513
total_digit3    0.8907543    0.9817450           NA    0.9959032    0.9393150
total_digit4    0.9102392    0.9766877    0.9959032           NA    0.9491674
total_digit5    0.9646054    0.9619513    0.9393150    0.9491674           NA

[1] "R^2 matrix for taxon: LORISIFORM"
             total_digit1 total_digit2 total_digit3 total_digit4 total_digit5
total_digit1           NA    0.5817319    0.7835986    0.9149144    0.8244058
total_digit2    0.5817319           NA    0.9060961    0.6307979    0.3938864
total_digit3    0.7835986    0.9060961           NA    0.8485835    0.6239906
total_digit4    0.9149144    0.6307979    0.8485835           NA    0.9098780
total_digit5    0.8244058    0.3938864    0.6239906    0.9098780           NA

[1] "R^2 matrix for taxon: LEMURID"
             total_digit1 total_digit2 total_digit3 total_digit4 total_digit5
total_digit1           NA    0.9513217    0.9244045    0.9532332    0.9753546
total_digit2    0.9513217           NA    0.9320955    0.9455698    0.9565207
total_digit3    0.9244045    0.9320955           NA    0.9879778    0.9401933
total_digit4    0.9532332    0.9455698    0.9879778           NA    0.9774962
total_digit5    0.9753546    0.9565207    0.9401933    0.9774962           NA

NA

1. For lorisiform:

• 5&2, 1&2, 5&3, 4&2 are least correlated.
• 1&4, 3&2, 5&4, 3&4, 5&1 are most correlated

2. For lemurid:

• Everything is pretty correlated –> probably because lemurid represents a large number of the data in its groupings
• 3&1, 3&2, 3&5 least correlated
• 4&3, 4&5 have highest correlation
• future plan: potentionally split by taxon next

3. For OWM:

• 1&3, 1&4, 1&2, 1&5 are least correlated
• everything else is pretty correlated (higher than .94)
• Odd pattern? Why are only digits 1 not correlated with anything?

Overall: - 4&5, 3&4, 1&5 have highest correlation - 3&5 least correlated

unique_taxa <- unique(autoclean$aagroup)

for (taxon in unique_taxa) {
  taxa_data <- autoclean[autoclean$aagroup == taxon, ]
  if (nrow(taxa_data) > 2 && all(complete.cases(taxa_data[, c("total_digit1", 
                                                              "total_digit2", 
                                                              "total_digit3", 
                                                              "total_digit4", 
                                                              "total_digit5")]))) {
  correlation_matrix <- cor(taxa_data[, c("total_digit1", "total_digit2", 
                                          "total_digit3", "total_digit4", 
                                          "total_digit5")])
  print(taxon)
  print(correlation_matrix)
  corrplot(correlation_matrix, method = "color", addCoef.col = "black", tl.col = 
             "black", tl.srt = 45,   title=taxon,    mar=c(0,0,1,0) )
  }
}

[1] "OWM"
             total_digit1 total_digit2 total_digit3 total_digit4 total_digit5
total_digit1    1.0000000    0.7677508    0.7426686    0.7516212    0.7718327
total_digit2    0.7677508    1.0000000    0.9804483    0.9799075    0.9720428
total_digit3    0.7426686    0.9804483    1.0000000    0.9980291    0.9905426
total_digit4    0.7516212    0.9799075    0.9980291    1.0000000    0.9940104
total_digit5    0.7718327    0.9720428    0.9905426    0.9940104    1.0000000
[1] "NWM"
             total_digit1 total_digit2 total_digit3 total_digit4 total_digit5
total_digit1    1.0000000    0.9821498    0.9761360    0.9641359    0.9593294
total_digit2    0.9821498    1.0000000    0.9922896    0.9792140    0.9739503
total_digit3    0.9761360    0.9922896    1.0000000    0.9931227    0.9838639
total_digit4    0.9641359    0.9792140    0.9931227    1.0000000    0.9941873
total_digit5    0.9593294    0.9739503    0.9838639    0.9941873    1.0000000

[1] "SCANDENTIA"
             total_digit1 total_digit2 total_digit3 total_digit4 total_digit5
total_digit1    1.0000000    0.9614239    0.9437978    0.9540646    0.9821433
total_digit2    0.9614239    1.0000000    0.9908305    0.9882751    0.9807912
total_digit3    0.9437978    0.9908305    1.0000000    0.9979495    0.9691826
total_digit4    0.9540646    0.9882751    0.9979495    1.0000000    0.9742522
total_digit5    0.9821433    0.9807912    0.9691826    0.9742522    1.0000000

[1] "LORISIFORM"
             total_digit1 total_digit2 total_digit3 total_digit4 total_digit5
total_digit1    1.0000000    0.7627135    0.8852110    0.9565116    0.9079680
total_digit2    0.7627135    1.0000000    0.9518908    0.7942279    0.6276037
total_digit3    0.8852110    0.9518908    1.0000000    0.9211859    0.7899308
total_digit4    0.9565116    0.7942279    0.9211859    1.0000000    0.9538753
total_digit5    0.9079680    0.6276037    0.7899308    0.9538753    1.0000000

[1] "LEMURID"
             total_digit1 total_digit2 total_digit3 total_digit4 total_digit5
total_digit1    1.0000000    0.9753572    0.9614595    0.9763366    0.9876004
total_digit2    0.9753572    1.0000000    0.9654509    0.9724041    0.9780188
total_digit3    0.9614595    0.9654509    1.0000000    0.9939707    0.9696357
total_digit4    0.9763366    0.9724041    0.9939707    1.0000000    0.9886841
total_digit5    0.9876004    0.9780188    0.9696357    0.9886841    1.0000000